2022/2023 Are going to be big years for point clouds in AEC. Collection has been approachable for awhile now, but digitizing is still a painful process due to the outdated software that's been an industry standard for awhile. I think that's likely to change in the next 18 months. Thanks for posting. I landed here while doing another lap around industry software companies to see what's new in their ability to work with point cloud (before I get fully invested in Microstation)
What makes you say that 22/23 will be big for pointclouds in AEC? We're well past the point of photogrammetry software maturing and becoming affordable; well past the point of inexpensive cameras being able to do the job. Based on the trends, what newly released or emerging technology do you believe will cause the rapid adoption at this point?
@@InspiredScience Lidar sensors and collection suites being more affordable and approachable, mainly. --Collecting-- Lidar and Photogrammetry is much, much easier now than it was 10 years ago, and at price points that small firms can stomach. The difficulty of fully utilizing the datasets has been in the post-processing/compilation software realm. It's the main barrier to entry right now. I think that 2023 is when we see better software solutions to make all of this more approachable. There's been software to address both for years, but they have high learning curves. The tipping point is when we have software for extracting data out of point clouds that's easy and intuitive to learn and use. Right now, becoming proficient in the established software suites is a career in itself.
@@pbsuasprogram2256 - I can see where that would be an intuitive view from someone without a lot of LiDAR development experience. I'm an actual software developer that's done work on LiDAR projects and while your statements would seem intuitive, they are not the reality, nor will LiDAR demonstrate a disproportionate rise in adoption for 22/23. The greatest rise in adoption at present are the SLAM-based systems where (to oversimplify a bit) the "post-processing" that you mention, is done in realtime. This is due primarily to price reductions in IMU's and in photogrammetry developments, which help orientate and align the LiDAR pointcloud. This is the driving force behind the $10k LiDAR units or DJI's new LiDAR devices. If you're limiting the conversation solely to close-range LiDAR for object-photogrammetry or AEC *interiors*, then certainly that will show a tremendous uptick in adoption solely because the very low-res/short-range LiDAR included in "Pro" IOS devices. However, please do not confuse that tech, or the price of that tech, with traditional expensive LiDAR. That tech has been relatively inexpensive for years. It's a limited scope, short-range, static LiDAR that is of little use in conventional pointcloud gathering (its original use is intended more as a more accurate sonar alternative and for all intents is an analog at a higher waveform). When you see SLAM-based LiDAR devices improve a bit (primarily via processing / mass-production), fall in price by about 80% or more, at that point you will see LiDAR adoption. That's about 3 years away, depending on market demand more than tech.
Yes, it's a great tool to have. Takes the photographer a couple hours to scan the house. Then maybe a day to get the MatterPak from Matterport. Then that's just about it. Maybe 10 minutes to bring the files into ReCap and save the .rcp.
I have some JPEG file got with point cloud data. Can you please guide me how to use them in project. If you have any video regarding this please drop a link here. This will be really helpfull.
If you haven't seen them, check out videos 1 & 2 from this series. When I get the raw files from my photographer there are a bunch of JPGs, but also a .obj, which I think is the important one and a .mtl which I think assigns the JPGs as materials in the point cloud. But I don't know a ton about how Recap works behind the scenes, it's always just worked for me.
We're looking to do a much larger building than a house for our work - basically a lodge/hotel. We've done the 3D scan and are having the vendor generate the point cloud (at which time we'll evaluate how good it is) My question is that for a big building, it'll be pretty time-consuming to retrace all the building elements - walls, floors, systems, etc - in Revit. While I certainly can do that from ground zero, are there any good ways to automate the process so that I'm not having to trace over literally everything that isn't furniture?
I don't know of an automated way to do that. There are online services that will do the as-built model for you, but that brings another set of issues and I personally would not trust that work...even if it's good too much effort to then modify to your own office standards. My thought on the point cloud workflow is that it's pretty much the same as before, just with a lot better notes. And 1/20th the time spent on site measuring.
I can't give away project files. But for learning purposes you are welcome to enroll in a course, on sale now while I transition my Hyperfine Revit content to a new brand called Revit Stuff: revit-stuff.teachable.com
Ben, you can do much better work with a higher-end photogrammetry tool (rather than using ReCap which sends everything over the 'net), even with a cheap camera. The matterport is truly dated and with a bit of knowledge/experience in photogrammetry, there are far better options at a lower cost.
@@HyperfineArchitecture For your workflow, I suspect the inexpensive/lower-end version of Metashape would probably produce the best results for the price. If you're not using it a great deal, Reality Capture has a per-use cost-structure and produces similar results. The beauty of Metashape is that in addition to producing some of the best results, it has some of the most powerful tools for anyone that is interested in investing a couple days to truly master all of the features. As far as a camera, any narrow-angle (for lack of a better term; non-wide-angle) camera that has a good IMU/GPS sensor that aids the software with accurate EXIF geo metadata. You can find some second-hand Samsung Android-based cameras that are Micro-4/3rds that are excellent due to the IMU/GPS and connectivity. You can also use any good phone that doesn't have too wide of an angle. The more popular the phone, the better, because it's more likely to have a pre-calibrated setting to compensate for lens distortion. That's another tip people often learn with experience; typically shoot in 4:3 even if you have a wider ration available, because in photogrammetry, the edge-distortion of lenses is a huge issue. Another tip people learn even quicker is that overlap is absolutely crucial. 85% or more sounds like overkill when you start; however, the more you do this, the more you prefer to over-shoot onsite, than to invest the time in post-processing/photogrammetry alignment. There are some good apps that will do this for you on Android/iPhones, without you having to press any shutter buttons, etc. A free option is Pix4D Capture (you won't be using the full-blown Pix4D membership/costly fee -- just the free app to acquire the photos). You simply walk around and use the phone like a video-camera (that's the best I can explain succinctly). The angle/camera movement is critical; practice in your own house and create true photogrammetry models to get it down. Using a gimbal with very little movement of the camera can be a big help because it helps with the consistency of the photos, allowing for better alignment in the photogrammetry software. I hope some of that can be of help :)
@@HyperfineArchitecture - it's not a drop in the bucket compared to what you do for others :). BTW - I noticed Pix4D has changed their nomenclature, the app you use for walking an interior without a UAV, would be: Pix4D "Catch" (vs Capture). You don't need to use an app like this, you can use any half-decent camera; however, the benefit of this sort of app is that it helps to automate/expedite the process a bit. If you get into this a bit and you run into issues, I'm always happy to help. Regardless of which approach/software you choose to go with, I've run most of them. From the oversimplified to the complex/expensive (eg Bentley Systems).
![ละลาย (LALALYE) - DAOU PITTAYA ต้าห์อู๋ พิทยา [OFFICIAL MV]](http://i.ytimg.com/vi/wo6r9TgausI/mqdefault.jpg)
2022/2023 Are going to be big years for point clouds in AEC. Collection has been approachable for awhile now, but digitizing is still a painful process due to the outdated software that's been an industry standard for awhile. I think that's likely to change in the next 18 months.
Thanks for posting. I landed here while doing another lap around industry software companies to see what's new in their ability to work with point cloud (before I get fully invested in Microstation)
What makes you say that 22/23 will be big for pointclouds in AEC? We're well past the point of photogrammetry software maturing and becoming affordable; well past the point of inexpensive cameras being able to do the job.
Based on the trends, what newly released or emerging technology do you believe will cause the rapid adoption at this point?
@@InspiredScience Lidar sensors and collection suites being more affordable and approachable, mainly. --Collecting-- Lidar and Photogrammetry is much, much easier now than it was 10 years ago, and at price points that small firms can stomach.
The difficulty of fully utilizing the datasets has been in the post-processing/compilation software realm. It's the main barrier to entry right now. I think that 2023 is when we see better software solutions to make all of this more approachable.
There's been software to address both for years, but they have high learning curves. The tipping point is when we have software for extracting data out of point clouds that's easy and intuitive to learn and use. Right now, becoming proficient in the established software suites is a career in itself.
@@pbsuasprogram2256 - I can see where that would be an intuitive view from someone without a lot of LiDAR development experience.
I'm an actual software developer that's done work on LiDAR projects and while your statements would seem intuitive, they are not the reality, nor will LiDAR demonstrate a disproportionate rise in adoption for 22/23.
The greatest rise in adoption at present are the SLAM-based systems where (to oversimplify a bit) the "post-processing" that you mention, is done in realtime. This is due primarily to price reductions in IMU's and in photogrammetry developments, which help orientate and align the LiDAR pointcloud.
This is the driving force behind the $10k LiDAR units or DJI's new LiDAR devices.
If you're limiting the conversation solely to close-range LiDAR for object-photogrammetry or AEC *interiors*, then certainly that will show a tremendous uptick in adoption solely because the very low-res/short-range LiDAR included in "Pro" IOS devices.
However, please do not confuse that tech, or the price of that tech, with traditional expensive LiDAR. That tech has been relatively inexpensive for years. It's a limited scope, short-range, static LiDAR that is of little use in conventional pointcloud gathering (its original use is intended more as a more accurate sonar alternative and for all intents is an analog at a higher waveform).
When you see SLAM-based LiDAR devices improve a bit (primarily via processing / mass-production), fall in price by about 80% or more, at that point you will see LiDAR adoption. That's about 3 years away, depending on market demand more than tech.
It’s insane how powerful this is! How long does it take to create the point cloud information?
Yes, it's a great tool to have. Takes the photographer a couple hours to scan the house. Then maybe a day to get the MatterPak from Matterport. Then that's just about it. Maybe 10 minutes to bring the files into ReCap and save the .rcp.
I have some JPEG file got with point cloud data. Can you please guide me how to use them in project. If you have any video regarding this please drop a link here. This will be really helpfull.
If you haven't seen them, check out videos 1 & 2 from this series. When I get the raw files from my photographer there are a bunch of JPGs, but also a .obj, which I think is the important one and a .mtl which I think assigns the JPGs as materials in the point cloud. But I don't know a ton about how Recap works behind the scenes, it's always just worked for me.
We're looking to do a much larger building than a house for our work - basically a lodge/hotel. We've done the 3D scan and are having the vendor generate the point cloud (at which time we'll evaluate how good it is) My question is that for a big building, it'll be pretty time-consuming to retrace all the building elements - walls, floors, systems, etc - in Revit. While I certainly can do that from ground zero, are there any good ways to automate the process so that I'm not having to trace over literally everything that isn't furniture?
I don't know of an automated way to do that. There are online services that will do the as-built model for you, but that brings another set of issues and I personally would not trust that work...even if it's good too much effort to then modify to your own office standards. My thought on the point cloud workflow is that it's pretty much the same as before, just with a lot better notes. And 1/20th the time spent on site measuring.
send it off to vietnam for modelling
Can I get this file for learning purpose please?
I can't give away project files. But for learning purposes you are welcome to enroll in a course, on sale now while I transition my Hyperfine Revit content to a new brand called Revit Stuff: revit-stuff.teachable.com
Ben, you can do much better work with a higher-end photogrammetry tool (rather than using ReCap which sends everything over the 'net), even with a cheap camera. The matterport is truly dated and with a bit of knowledge/experience in photogrammetry, there are far better options at a lower cost.
What do you suggest I use?
@@HyperfineArchitecture For your workflow, I suspect the inexpensive/lower-end version of Metashape would probably produce the best results for the price.
If you're not using it a great deal, Reality Capture has a per-use cost-structure and produces similar results.
The beauty of Metashape is that in addition to producing some of the best results, it has some of the most powerful tools for anyone that is interested in investing a couple days to truly master all of the features.
As far as a camera, any narrow-angle (for lack of a better term; non-wide-angle) camera that has a good IMU/GPS sensor that aids the software with accurate EXIF geo metadata. You can find some second-hand Samsung Android-based cameras that are Micro-4/3rds that are excellent due to the IMU/GPS and connectivity.
You can also use any good phone that doesn't have too wide of an angle. The more popular the phone, the better, because it's more likely to have a pre-calibrated setting to compensate for lens distortion.
That's another tip people often learn with experience; typically shoot in 4:3 even if you have a wider ration available, because in photogrammetry, the edge-distortion of lenses is a huge issue. Another tip people learn even quicker is that overlap is absolutely crucial. 85% or more sounds like overkill when you start; however, the more you do this, the more you prefer to over-shoot onsite, than to invest the time in post-processing/photogrammetry alignment.
There are some good apps that will do this for you on Android/iPhones, without you having to press any shutter buttons, etc. A free option is Pix4D Capture (you won't be using the full-blown Pix4D membership/costly fee -- just the free app to acquire the photos). You simply walk around and use the phone like a video-camera (that's the best I can explain succinctly).
The angle/camera movement is critical; practice in your own house and create true photogrammetry models to get it down. Using a gimbal with very little movement of the camera can be a big help because it helps with the consistency of the photos, allowing for better alignment in the photogrammetry software.
I hope some of that can be of help :)
That’s a hell of a response! Thanks for the info.
@@HyperfineArchitecture - it's not a drop in the bucket compared to what you do for others :).
BTW - I noticed Pix4D has changed their nomenclature, the app you use for walking an interior without a UAV, would be: Pix4D "Catch" (vs Capture).
You don't need to use an app like this, you can use any half-decent camera; however, the benefit of this sort of app is that it helps to automate/expedite the process a bit.
If you get into this a bit and you run into issues, I'm always happy to help. Regardless of which approach/software you choose to go with, I've run most of them. From the oversimplified to the complex/expensive (eg Bentley Systems).